The Discovery of a Hidden Broad Line AGN in a Bulgeless Galaxy: Keck NIR Spectroscopic Observations of SDSS J085153.64+392611.76
Thomas Bohn, Gabriela Canalizo, Shobita Satyapal, Ryan W. Pfeifle
DDraft version July 29, 2020
Typeset using L A TEX twocolumn style in AASTeX62
The Discovery of a Hidden Broad-line AGN in a Bulgeless Galaxy: Keck NIR Spectroscopic Observations of SDSSJ085153.64+392611.76
Thomas Bohn, Gabriela Canalizo, Shobita Satyapal, and Ryan W. Pfeifle University of California, Riverside, Department of Physics & Astronomy900 University Ave., Riverside, CA 92521, USA George Mason University, Department of Physics & AstronomyMS3F3, 4400 University Drive, Fairfax, VA 22030, USA
Submitted to ApJABSTRACTWe report the discovery of a buried, active supermassive black hole (SMBH) in SDSSJ085153.64+392611.76, a bulgeless Seyfert 2 (Sy2) galaxy. Keck near-infrared observations reveal ahidden broad-line region, allowing for the rare case where strong constraints can be placed on both theBH mass and bulge component. Using virial mass estimators, we obtain a BH mass of log( M BH /M (cid:12) )= 6 . ± .
50. This is one of the only Sy2 active galactic nuclei (AGN) hosted in a bulgeless galaxy witha virial BH mass estimate and could provide important constraints on the formation scenarios of theBH seed population. The lack of a bulge component suggests that the SMBH has grown quiescently,likely caused by secular processes independent of major mergers. In the absence of a detectable bulgecomponent, we find the M BH – M stellar relation to be more reliable than the M BH – M bulge relation. Inaddition, we detect extended narrow Pa α emission that allows us to create a rotation curve wherewe see counterrotating gas within the central kiloparsec (kpc). Possible causes of this counterrotationinclude a galactic bar or disruption of the inner gas by a recent fly-by of a companion galaxy. This inturn could have triggered accretion onto the central SMBH in the current AGN phase. Keywords: galaxies: active — galaxies: bulges — galaxies: evolution — galaxies: Seyfert — infrared:galaxies INTRODUCTIONThe advent of the discovery that supermassive blackholes (SMBHs) lie at the center of virtually all massivegalaxies has promoted the idea that these black holes(BHs) play a fundamental role in galaxy formation andevolution (for a review, see Kormendy & Ho 2013). Well-known relations such as BH mass ( M BH ) correlatingwith stellar velocity dispersion, M BH – σ , (eg., Ferrarese& Merritt 2000; Gebhardt et al. 2000; McConnell & Ma2013), and stellar bulge mass, M BH – M bulge , (eg., Mar-coni & Hunt 2003; H¨aring & Rix 2004), provide a picturethat BH growth accompanies central bulge growth. Anoften-suggested scenario of this interaction involves ma-jor mergers that not only fuel BH growth but can alsotrigger the buildup of the bulge component (eg., Kauff- [email protected] mann et al. 2003; Di Matteo et al. 2005; Ellison et al.2011). Subsequent feedback from the accreting BH (ac-tive galactic nucleus [AGN]) can help quench star for-mation by either expelling gas out of the galaxy (eg.,Kauffmann & Haehnelt 2000) or by heating the gas inthe halo and preventing it from feeding the disk (eg.,Bower et al. 2006; Croton et al. 2006). As a result,the galaxy evolves toward the well-defined red sequence.Therefore, galaxy growth has long been thought to behierarchical, with major mergers providing the necessaryconditions for BHs and their host galaxies to reach theirobserved masses (eg., Sanders et al. 1988; Kauffmannet al. 1993). In addition to major mergers affecting BHgrowth, stochastic fueling from dense gas clouds reach-ing the nucleus can also trigger AGN at the low/mid-luminosity regime (Hopkins et al. 2014).This scenario of BH growth accompanying bulge de-velopment through mergers highlights the importanceof studying the secular evolution of BHs in galaxies a r X i v : . [ a s t r o - ph . GA ] J u l Bohn et al. that have had a quiescent merger history. Unlike bulge-dominated galaxies whose BHs have had accelerated ac-cretion, BHs in likely merger-free galaxies (such as bul-geless galaxies) have grown largely independent of ma-jor interactions. Therefore, the mass distribution andoccupation fraction of these BHs can provide impor-tant clues to the original seed population and seculartriggering mechanisms. Additionally, current BH scal-ing relations lack significant contributions from bulgelessgalaxies that can misrepresent the AGN population asa whole. Studying these quiescently grown BHs is thuscritical to our understanding of BH growth and theircontribution to their host galaxy evolution.Discoveries of disk-dominated (low bulge-to-total lightratio, B/T) and bulgeless galaxies hosting low- tointermediate-luminosity AGN have been limited, butrecent estimates of their BH masses (eg., Filippenko &Ho 2003; Satyapal et al. 2007; McAlpine et al. 2011;Secrest et al. 2012; Reines et al. 2013; Simmons et al.2013, 2017) indicate that they can be up to 10 M (cid:12) .These findings are starting to show that a central bulgeis not a requirement to have a SMBH and that M BH in disk-dominated galaxies are likely correlated to thetotal stellar mass of the galaxy ( M stellar ) rather thanto the mass of the bulge, M bulge (Simmons et al. 2017;Martin et al. 2018). In addition, coevolution of BHsand galaxies through merger-free processes, such as diskinstabilities and secular growth, has been previouslysuggested (eg., Kormendy & Kennicutt 2004; Greeneet al. 2010; Schawinski et al. 2011b), and these pro-cesses may be able to grow the central BHs to theirtypical observed masses (Simmons et al. 2013; Martinet al. 2018). This suggests that AGN feedback or per-haps some broader, galaxy-wide process regulates theamount of matter that the BH is allowed to accrete.However, the number of purely bulgeless galaxies with M BH estimates in the literature represents a very smallfraction of the total bulgeless population, and it is verylikely that optical catalogs misidentify or exclude deeplyburied AGN in dusty, late-type galaxies. Additionalproblems arise in verifying the true morphology of thecentral region, and it is often difficult to rule out thepresence of small bulges. This is particularly problem-atic for Seyfert 1 (Sy1) galaxies, where the bright AGNis in our direct line of sight, compromising the reliabil-ity of bulge–disk decompositions employed to measurethe total bulge component. While the visible broad-line region (BLR) in these galaxies allows us to obtainestimates of M BH through the virial method, the lightfrom the AGN can preclude us from detecting a smallbulge component. Sy2 galaxies, on the other hand, allowfor much more stringent constraints on the presence of bulges, since their AGN is hidden from our line of sight.However, for the same reason, M BH estimates are moredifficult to come by. Several methods have been used inan attempt to detect the ‘hidden’ BLR in Sy2 galaxies,including spectropolarimetry (Antonucci & Miller 1985)and high S/N near-infrared (NIR) observations whereextinction is less severe (Veilleux et al. 1997; Lampertiet al. 2017). These studies have revealed that only 10-20% of Sy2 galaxies show a BLR in the NIR, likely dueto strong obscuration.In this article, we present the discovery of a hid-den, NIR BLR found in J085153.64+392611.76, here-after J0851+3926, a spiral galaxy at redshift 0.1296 thatshows no signs of a bulge component and is part of alarger study of bulgeless galaxies (T. Bohn et al., inpreparation; Fig. 1). J0851+3926 is listed as ‘starform-ing’ under the Subclass keyword by the Sloan DigitalSky Survey (SDSS), and the SDSS spectrum, althoughshowing composite narrow-line ratios, does not showclear broad Balmer lines. Since there is no AGN con-tribution at the center, we can put strong constraintson the presence of a bulge using optical photometry.This allows for the rare chance of obtaining a robustBH mass estimate from the NIR broad line while alsoputting strong constraints on any possible bulge compo-nent. In Section 2, we describe the construction of oursample, observations, and reduction procedure. Section3 presents the results of surface brightness decomposi-tions, the BH mass, intrinsic extinction, and observedgas dynamics. Section 4 compares J0851+3926 to otherbulgeless galaxies and M BH –galaxy relations. Addition-ally, we discuss possible triggering mechanisms of theAGN. We adopt a standard ΛCDM cosmology with H = 70 km s − Mpc − , Ω M = 0.3, and Ω Λ = 0.7. DATA AND OBSERVATIONS2.1.
Data Selection
Since large optical surveys can miss deeply buriedAGN, our selection process focuses on infrared (IR) se-lection techniques. Satyapal et al. (2014), hereafter S14,selected galaxies believed to host obscured AGN usingmid-infrared colors and the AGN selection criteria pre-sented in Jarrett et al. (2011). S14 suggested that IRindicators could be used to identify optically obscuredAGN based on their strong IR colors that separate themfrom stellar processes. Motivated by these findings, ourselection process followed closely to that of S14. To sum-marize, we formed an initial sample of bulgeless galaxiesby drawing from Simard et al. (2011), who performedbulge-disk decompositions using
GIM2D (Simard et al.2002) of 1.12 million galaxies from SDSS DR7. Thesurface brightness, point-spread function (PSF) con- idden BLR in Bulgeless Galaxy log ([NII]/H ) l o g ([ O III ] / H ) Star-forming Composite AGN
S14J0851+3926 W W W W AGN Demarcation Box S14J0851
Figure 1.
BPT (left) and WISE color (right) plots of the bulgeless sample. Red diamonds and the blue star represent thesample selected by S14 and J0851+3926, respectively. The inclusion contours are drawn at σ intervals (68%, 95%, and 99.5%).The lines separating the AGN (Kewley et al. 2001) and composite (Kauffmann et al. 2003) regions from the starforming in theBPT diagram are shown as solid and dashed lines, respectively. The AGN demarcation region is shown as defined in Jarrettet al. (2011). Note that J0851+3926 falls in the composite region and is on the border of the AGN demarcation box. volved bulge–disk decompositions were done in bothSDSS r and g bands. Three different galaxy fittingmodels were utilized: a bulge ( n b = 4) + disk model,a free-floating S´ersic index “bulge” ( n b = free) + diskmodel, and a pure S´ersic model. We used the modelwith a free-floating bulge index in order to select galax-ies with a B/T equal to 0.00. Of the 632,952 galaxieswithin a redshift of z < .
2, only 19,136 have B/T =0.00 in both r and g bands. Using fluxes taken fromthe Portsmouth spectroscopic reanalysis emissionLine-sPort table (Thomas et al. 2013) in SDSS, we con-structed a Baldwin, Phillips, & Terlevich (BPT) dia-gram (Baldwin et al. 1981) using [O III ] λ β and[N II ] λ α line ratios (see Figure 1, left panel).The Portsmouth analysis accounted for stellar absorp-tion features by using the Gas AND Absorption LineFitting ( GANDALF ; Sarzi et al. (2006)) and the pe-nalized Pixel Fitting ( pPXF ; Cappellari & Emsellem2004) routines. By fitting for the stellar absorption fea-tures, the fluxes of the hydrogen lines, specifically H α and H β , increase, causing galaxies to generally shift to-ward the lower left (i.e., the starforming region) of theBPT diagram. We excluded the 648 galaxies with noregistered Portsmouth fluxes, which left only 18,488 bul-geless galaxies in our sample. Following the AGN clas-sification scheme presented in Kewley et al. (2001), only143 (0.77%) galaxies are identified as AGN. However, https://gandalfcode.github.io ∼ mxc/software/ many spectra can contain contributions from both theAGN and star-forming H II regions. As a result, galax-ies hosting relatively weak AGN can fall below this line.Kauffmann et al. (2003) defined a composite region be-tween the AGN and star-forming portions of the dia-gram, where 950 (5.14%) galaxies of our sample fall.Galaxies in the composite region are generally believedto have a mixture of AGN and star-forming emission.However, merger-driven shocks can reproduce AGN-HII emission-line ratios (Rich et al. 2014), so galaxiesthat fall in this region cannot be definitively classifiedas AGN without other lines of evidence.AGN at low redshift should be considerably redderthan inactive galaxies (Stern et al. 2012; Assef et al.2013). Utilizing the Wide-field Infrared Survey Explorer(WISE) All-Sky Data Release (Wright et al. 2010), weobtained WISE band magnitudes W µ m), W µ m), and W µ m) matched within 1 (cid:48)(cid:48) for our SDSSbulgeless sample. Of the 18,488 SDSS bulgeless galax-ies, 18,146 (98.15%) have registered WISE magnitudesin the required bands. In Figure 1 (right panel), weemployed the WISE color diagnostic presented in Jar-rett et al. (2011). Here, they define a demarcation zoneseparating AGNs using W W W W W W > Bohn et al.
Table 1.
Observation LogInstrument Date Seeing Exp. Time PA Extraction Aperture Air Mass Telluric(YYYY-mm-dd) (arcsec) (s) (deg) (arcsec)NIRSPEC 2018 March 5 ∼ a
46 1.34 1.06 HD 63610NIRES 2019 Mar 25 ∼ b
94 1.33 1.07 HD 63610 a ×
240 s exposures (two ABBA sets) were taken. b ×
240 s exposures were taken.
NIR Observations and Reductions
NIR spectroscopy of J0851+3926 was obtained on twoseparate dates: on 2018 March 5 using Keck II NIR-SPEC (McLean et al. 1998), and on 2019 March 25with Keck II NIRES (Wilson et al. 2004). NIRSPECis a NIR echelle spectrograph with a wavelength cover-age from 0.9 to 5.5 µ m. The NIRSPEC-7 filter was usedin low-resolution mode with a cross-dispersion angle of35.38 ◦ . This resulted in a wavelength coverage of ∼ µ m. The 42 (cid:48)(cid:48) × . (cid:48)(cid:48) slit was used, and a spectralresolution of ∼
120 km s − at the observed wavelengthof Pa α was measured with a seeing of ∼ (cid:48)(cid:48) . Obser-vations throughout the night were done under variableand heavy cloud cover. While telluric and flux stan-dards were observed before and after the science object,the amount of extinction was highly variable, and thusthe flux calibration for these data is uncertain. Notethat these observations were done before the NIRSPECupgrade. NIRES is a NIR echelette spectrograph and ithas a fixed configuration. The single slit is 18 (cid:48)(cid:48) × . (cid:48)(cid:48) and the wavelength coverage is set from 0.94 to 2.45 µ m across five orders. There is a small gap in cover-age between 1.85 and 1.88 µ m, but this is a region oflow atmospheric transmission. The spectral resolutionat Pa α was ∼
85 km s − and the seeing was typically ∼ (cid:48)(cid:48) throughout the night. Observations were doneunder mostly clear conditions, and so the majority of theanalysis was done with the NIRES data. Individual ex-posures for both sets of observations were 4 minutes eachand were done using the standard ABBA nodding. AnA0 telluric standard star (with measured magnitudes in K , H , and J bands) was observed either directly beforeor after the target galaxy to correct for the atmosphericabsorption features. The total exposure times for NIR-SPEC and NIRES were 32 and 20 minutes, respectively.A summary of the NIR observations is shown in Table1. The data were reduced using two modified pipelines.The first provided flat-fielding and a robust backgroundsubtraction by using techniques described in Kelson (2003) and Becker et al. (2009). In short, this routinemaps the 2D science frame and models the sky back-ground before rectification, thus reducing the possibilityof artifacts appearing due to the binning of sharp fea-tures. The sky subtraction attained with this procedureis excellent, despite the strong OH lines present in theNIR; the procedure is also quite insensitive to cosmicrays and hot pixels and is reliable regardless of skylineintensity.Rectification, telluric correction, wavelength calibra-tion, and extraction were all done with a slightly modi-fied version of REDSPEC . The sizes of the extractedaperture are listed in Table 1. The 1D spectrum werethen median combined. Flux calibration was done us-ing the telluric star and the Spitzer Science Center unitconverter to convert the magnitude of the star to theassociated flux in that band. A small corrective factor( < X-Ray Observations and Reductions
J0851+3926 was observed for 19.8 ks with the ACIS-S instrument on board the
Chandra
X-ray Observatoryon 2020 January 19, with the target centered at theaim point of the ACIS-S3 chip. The data were reducedand analyzed using the Chandra Interactive Analysis ofObservations ( ciao ) software package (Fruscione et al.2006) version 4.11 along with version 4.8.2 of the Cali-bration Database ( caldb ). A circular aperture of 1.5 (cid:48)(cid:48) in radius was centered on the coordinates of the galaxynucleus, from which source counts were extracted. Thebackground counts were extracted using a circular aper-ture of radius 25 (cid:48)(cid:48) and was placed in a nearby area free ofother sources. Full (0.3 – 8 keV), soft (0.3 – 2 keV), andhard (2 – 8 keV) counts were extracted from energy- http://ssc.spitzer.caltech.edu/warmmission/propkit/pet/magtojy/ idden BLR in Bulgeless Galaxy dmextract package in ciao and error bounds were calculated using Gehrelsstatistics (Gehrels 1986). ANALYSIS3.1.
GALFIT
Fitting
J0851+3926 is a Sy2 galaxy with no visible AGN tosaturate or blend with a possible bulge. In order to placestringent constraints on the presence of a small bulge, weperformed two-dimensional decompositions using
GAL-FIT (Peng et al. 2002, 2010) and ran fits using variouscombinations of PSF and S´ersic profiles. The PSF wasconstructed from the psField file provided by SDSS andhad a full-width half-maximum (FWHM) of 0.717 (cid:48)(cid:48) (1.65kpc at the redshift of J0851+3926). Due to the lack ofa resolved core component, GALFIT could never con-verge on a reasonable solution when a PSF or two S´ersicprofiles were used. As expected for a bulgeless galaxy,
GALFIT could only properly converge on a solutioncontaining a single S´ersic profile. As a check, we usedvarious initial index values reflecting a de Vaucouleurbulge ( n = 4), a pseudobulge ( n ∼ n = 1). Regardless of the initial valuesused, GALFIT consistently converged on an index of n = 0 . (cid:48)(cid:48) , where the galaxy ends and the backgroundnoise starts to dominate. However, there are some smalloscillations in the residuals that are due to spiral arms inthe disk of the galaxy, as seen in the 2D residual image.These spiral arms could alter the results of the fitting,particularly in the central region. To factor these out,we tried fitting them using a combination of a powerfunction and Fourier modes, but the resolution and sur-face brightness of the arms were too low for GALFIT to converge on any solution. We subsequently created amask using the residuals from the fit and left the cen-tral region unmasked. Fitting the galaxy with the maskgreatly reduced the residuals at the center and gave aS´ersic index of n = 0 .
89 (see Figure 2b). The centralunmasked region within roughly 1.5 (cid:48)(cid:48) is almost perfectlydescribed as a disk with no bulge component.Although the fits in Figure 2 match the data quitewell, the possibility of an unresolved bulge or pseudob-ulge component cannot be dismissed. We obtained an https://users.obs.carnegiescience.edu/peng/work/galfit/galfit.html upper limit to the bulge mass by estimating the magni-tude of a PSF (i.e., an unresolved component) that canaccount for the residuals closest to the core in Figure 2a(i.e., the fit without masking the spiral arms). Forcinga PSF that is about 6.5 fainter in magnitude than thetotal galaxy removed all traces of the central residualsof the original fit. Any PSF brighter than this resultsin an oversubtraction. Thus, we take this PSF to be astrong upper limit to the light contribution by an unre-solved bulge. This results in a B/T ≤ M stellar from Chang et al.(2015), who provide a catalog of stellar masses usingSDSS and WISE photometry. Here, SED fitting wasperformed using both optical and IR imaging to obtainstellar masses. For J0851+3926, a total stellar mass oflog( M stellar /M (cid:12) ) = 10.61 ± M bulge /M (cid:12) ) ≤ BH Mass of J0851+3926
Both the NIRES and NIRSPEC spectra of J0851+3926clearly show a broad and a narrow component of Pa α (see Figs. 3 and 8). While broad Pa α is certainly in-dicative of AGN activity, Baldassare et al. (2016) andother follow-up studies of AGN candidates with broademission have shown that supernovae (SNe) and otherstellar activity can produce similar broad features. TypeII SNe (Pritchard et al. 2012) and luminous blue vari-ables (Smith et al. 2011) are known to produce broadrecombination lines up to thousands of kilometers persecond. If the broad Pa α observed in J0851+3926 werepowered by a SN, the broad emission would have per-sisted for more than 380 days based on the observationdates of our two sets of spectra (see Table 1), and thisin turn would indicate that the SN would likely be aType II-P. Using this time scale, we would expect to seeother NIR SN features such as O I , Mg I , and Ca I (e.g.,Rho et al. 2018). However, we do not see any of thesefeatures in either of our NIR observations. Additionally,we would expect the line profile to change significantlyover this time (Rho et al. 2018) but our two measure-ments of the broad Pa α width are consistent with eachother, 1489 (NIRSPEC) and 1363 (NIRES) km s − (seeTable 2).Another potential origin of a broad line could bepowerful outflows powered by star formation. Broad, Bohn et al.
SDSS J085153.64+392611.7
Best-fit Model
Residuals S u r f a c e B r i g h t n e ss , µ [ m a g a r c s e c − ] DataModelDisk
Semi-major Axis [arcsec] ∆ µ (a) SDSS J085153.64+392611.7
Best-fit Model
Residuals S u r f a c e B r i g h t n e ss , µ [ m a g a r c s e c − ] DataModelDisk
Semi-major Axis [arcsec] ∆ µ (b) Figure 2.
GALFIT decomposition fits to the SDSS image of J0851+3926. Panel (a) is without the outer spirals masked anduses a free S´ersic index, while panel (b) has the spirals masked and the S´ersic is fixed at n = 1. In each figure, the fits of themodel and disk to the data (black dotted line) are shown in the top left panel as blue and yellow lines, respectively. The modelincludes both the disk and sky background component. The bottom left panel shows the residuals to the model fit. The shadedgray areas represent the 1- σ error. The three right panels show postage stamps of the raw SDSS image, best-fit model, andresiduals. idden BLR in Bulgeless Galaxy III ] λ α in J0851+3926 is powered by an outflow,we fit the optical lines in the SDSS spectrum by usingBayesian AGN Decomposition Analysis for SDSS Spec-tra ( BADASS ; R. Sexton et al. 2020, in preparation),a spectral analysis tool where we included fits to thestellar and Fe II features, as well as multiple compo-nents to emission lines. The code allows the user totest for the presence of outflows by setting various con-straints on parameters such as amplitude, width, andvelocity offset. All reasonable criteria came back neg-ative for outflows, so we forced a blueshifted, outflowcomponent to the [O III ] λ α is the BLR of an AGN.As such, we can use this broad line to estimate BH massusing the virial method.To obtain a BH mass, a common and reliable methodis through the virial relation, defined as M BH = f V RG (1)where f is the virial coefficient; V is the velocity ofthe broad-line gas that is responding to the continuumvariations; R is the distance from the broad emission gasto the central continuum source and is equal to cτ where τ is the time delay and c is the speed of light; and G is the gravitational constant. The FWHM of the broademission line, typically seen in H α or H β in the optical,can be used for the value of V . Here, the width of thebroad line stems from the Doppler effect of the gas inthe accretion disk revolving around the BH. The value of R is estimated empirically using the optical luminosityof the AGN as a proxy (Kaspi et al. 2005; Bentz et al.2013).The spectra were fit using emcee , an affine-invariantMarkov Chain Monte Carlo (MCMC) ensemble sampler(Foreman-Mackey et al. 2013). A broad component anda narrow component, along with a second-order poly-nomial for the continuum, were fit simultaneously (see https://github.com/remingtonsexton/BADASS2 Figure 3). Both the narrow and broad components weretreated as Gaussians with amplitude, FWHM, and offsetfrom rest-frame wavelength as free variables. The BHmass was obtained following the estimators presented byKim et al. (2018), where they adopted the virial factorlog f = 0.05 ± MM (cid:12) = 10 . ± . (cid:18) L Pa α erg s − (cid:19) . ± . (cid:18) FWHM Pa α km s − (cid:19) (2)where the FWHM of broad Pa α is the analog to thevelocity in the virial mass estimator and L Pa α of thebroad component is the analog to the distance to theBLR.Our NIRSPEC observations were done under heavycloud cover, making it difficult to accurately estimatethe degree of extinction. Thus, the flux and BH valuesare unreliable, but we leave them listed in Table 2 forcompleteness and show the fit to the spectrum in Ap-pendix A. For the remainder of the article, we will onlyuse our NIRES data for analysis. From Equation (2),our NIRES measurements give a BH mass of (4 . +1 . − . ) × M (cid:12) (see Table 2 for relevant values), where thecalculated uncertainties come from the random error es-timates. Accounting for systematic uncertainties, virialBH mass estimates typically have errors of 0.4 - 0.5 dex(eg., Shen 2013; Reines & Volonteri 2015). We adopta conservative error estimate of 0.5 dex. For a detaileddescription on virial mass uncertainties, see Sexton et al.(2019). 3.3. Extinction
The presence of broad Pa α and the lack of strongbroad lines in the optical imply that there is heavyobscuration present. In order to quantify the extinc-tion toward the BLR, we measured broad hydrogenemission-line ratios and assumed a Cardelli reddeninglaw (Cardelli et al. 1989) with an extinction factor R V = 3.1. While the observed Pa β wavelength fell in aregion of strong atmospheric absorption, Pa γ emissionis observable in the J band, which allowed us to ob-tain an upper limit to the broad Pa γ flux. We alsofit the optical SDSS spectrum using BADASS , wherefits to the stellar and Fe II features were included (seeR. Sexton et al. 2020, in preparation for further de-tails). Two different models were used to fit the data:one excluding broad components (i.e., only narrow com-ponents) and the other including broad components (seeFig. 9 in Appendix B). No broad H β could be properlyfit, but the code did converge on a solution to broadH α . We compared the fits to H α using the F -test: F = Bohn et al.
Table 2. Pa α MeasurementsInstrument Flux
Broad
Flux
Broad (Ext.) a FWHM
Broad M BH M BH (Ext.) a (10 − erg cm − s − ˚A − ) (km s − ) (10 M (cid:12) )NIRSPEC b +0 . − . +0 . − . ±
184 2.90 +2 . − . +3 . − . NIRES 1.08 +0 . − . +0 . . ±
31 4.47 +1 . − . +2 . − . Note —Listed errors are from random errors in the fitting process. a Extinction-corrected values. b NIRSPEC measurements are ignored owing to the high extinction caused by heavy cloud cover. F ( e r g c m s Å ) SpectrumBest FitPa Narrow ComponentPa Broad Component
Rest Wavelength (Angstroms) F R = 0.89 Figure 3.
The MCMC fit to the NIRES spectrum. In the top panel, Pa α is centered with the best fit plotted over the spectrum.Below the spectrum are the narrow (dotted) and broad (solid) components. The bottom panel plots the residuals, the 1 σ noiselevel (horizontal dotted lines), and the R value. ( σ single ) / ( σ double ) , where σ is the standard deviationof the residuals using either single or double gaussiancomponents, for which we obtain F = 1.41. Based onthis, we cannot say for certain whether adding a broadcomponent is justifiable. A value closer to 2 or 3 wouldprovide convincing evidence that a broad componentshould be fit. Although the fits do suggest that somebroad emission is present, deeper observations will beneeded to clear the ambiguity of the broad H α emission.The observed line ratios from the fits are Pa α /Pa γ ≥ α /H α ≥ α /Pa γ = 3.22 and Pa α /H α = 0.10, were obtained from Dopita& Sutherland (2003), where we assumed an electron den-sity of n e = 10 cm − and temperature T e = 15,000 K.Using the Cardelli reddening law, we estimated an ex-tinction of E Pa γ ( B − V ) ≥ E H α ( B − V ) ≥ E H α ( B − V ) ≥ +2 . − . × M (cid:12) , which we will use forthe remainder of the article (see Table 2 for extinction-corrected values).This high degree of extinction could explain the lackof other AGN indicators, such as [Si VI ]1.963 µ m and idden BLR in Bulgeless Galaxy VI ] emission based on the ob-served WISE W µ m) flux. The objects presentedby M¨uller-S´anchez et al. (2018) appear to follow a re-lation between [Si VI ] and W VI ]) =0.74 × log( W
2) - 6.6045 (J. Cann, private communica-tion). After adjusting for extinction, the expected [Si VI ]flux is 1 . × − erg cm − s − . In order to determinewhether if this could be detected, we estimate the flux ofa gaussian with an amplitude of 1 σ of the noise level anda width of the resolution element of the telescope. Thisresulted in a flux of 1 . × − erg cm − s − . Thus,any [Si VI ] emission will be at most comparable to thenoise level. Since [Si VI ] is one of the most prominentcoronal lines in the NIR, we also do not expect to seeother coronal line features.3.4. X-Ray Observations
The absorbing hydrogen column density ( N H ) of Sy2galaxies is expected to be high, on the order of ∼ cm − (Jaffarian & Gaskell 2020). This is consis-tent with the unified model, where the active nuclei inSy2 galaxies are believed to be heavily obscured dueto orientation effects. Coupled with the high-extinctionestimates calculated in Section 3.3, it is not surprisingthat we did not detect any statistically significant X-rayemission in the Chandra observations. We calculateda 3 σ upper limit on the counts, ∼
6, and assuming apower-law index of 1.8, the upper limit to the hard X-ray 2 – 10 keV luminosity, L −
10 keV was estimated to be1 . × erg s − . Using Equation (1) from R. W. Pfei-fle et al. (2020, in preparation), a column density can beestimated from L −
10 keV and the WISE 12 µ m luminos-ity ( L µ m ). We found a lower limit line-of-sight columndensity of log(N H / cm − ) ≥ .
43, which suggests thatthe obscuring region is Compton thick. This estimateof N H along with the results presented in Jaffarian &Gaskell (2020) implies an E ( B − V ) > Rotation Curve
Visible in both the NIRSPEC and NIRES 2D spectrais extended narrow Pa α emission that traces the gasin the disk out to about 8 kpc. The spatially resolvednarrow emission allows us to construct a rotation curve.Plotted in Figure 4 are two rotation curves based ondifferent slit orientations, one along the semi-major axis(top) and the other oriented almost perpendicular tothat (bottom). For each curve, velocities were measured in either ∼ ∼ emcee routine as wasused in Section 3.2. The inclination angle is taken fromthe output of GALFIT (see Figure 2). Plotted in grayis the expected velocity curve using a Navarro-Frenk-White (NFW) dark matter density profile (Navarro et al.1996). The width of the curve arises by varying theconcentration parameter from 8 to 15, with the dottedline representing a value of 10.As shown in the top panel of Figure 4, there appearsto be counterrotating gas within the central kpc. Thiscounterrotation is not apparent when the slit is ori-ented +60 ◦ (bottom panel). The limited extension ofthe counter-rotation and the orientation of the slit alongthe central component of the spiral could suggest that abar is causing the velocity disruption. Another scenariothat could explain this counterrotation is a fly-by of apossible companion galaxy. We discuss these scenariosfurther in Section 4.3. DISCUSSIONThe virial method offers one of the most reliable meth-ods of estimating BH masses, and J0851+3928 is one ofonly a handful of known bulgeless galaxies for which BHmass can be estimated by this method. In the followingsections, we first compare the M BH of J0851+3928 tothose of other bulgeless galaxies, followed by a compar-ison to a much broader sample that includes all mor-phological types. We compare M BH to both the galacticbulge mass ( M bulge ) and total stellar mass of the galaxy( M stellar ). Note that the estimates for M BH used herecome from methods using the gravitational potential ofthe SMBH (viral mass estimators) or those using theAGN as the flux source (X-ray estimates). We refrainfrom using BH masses derived from relations based ongalaxy properties, including M BH - σ and M BH - φ (spiralarm pitch angle).4.1. Comparisons with Other Bulgeless Galaxies
One of the first examples of an AGN in a bulgelessgalaxy in the literature is NGC 4395, a nearby Sy1galaxy hosting an intermediate-mass BH (IMBH; Filip-penko & Sargent 1989; Filippenko & Ho 2003). Ultravi-olet reverberation mapping has estimated the BH massto be (3 . ± . × M (cid:12) (Peterson et al. 2005) andthis has been verified by subsequent direct dynamicalmass measurements (den Brok et al. 2015). Jiang et al.(2011) report seven broad-line AGN (only 5% of theirsample) where a pure exponential disk provided the bestfit, indicating the lack of a bulge component. BH masses0 Bohn et al.
Distance from center (kpc) V e l o c i t y ( k m s − ) NE (a) Distance from center (kpc) V e l o c i t y ( k m s − ) (b) Figure 4.
Rotation curves of J0851+3926. A NFW profile is plotted for reference in the left panels. The shaded gray regionrepresents the concentration parameter varying from 8 to 15, with 10 represented as the dotted line. The extended narrow-linevelocities of Pa α , blue for approaching and red for receding gas, are plotted against their distance from the center in kpc. Onthe right, the orientation of the slit is shown as a white strip, while the colored lines correspond to the direction of the gas andthe distance to which the extended emission is detectable. in this sample range from 10 . to 10 . M (cid:12) . As notedby the authors, four of these have bar structures andthe bright AGN at the center could hide a small bulge.Simmons et al. (2017) provide a large sample (101 galax-ies) of type 1 AGN in disk-dominated galaxies with BHmasses ranging from 2 × to 9 × M (cid:12) . However,the mean B/T of this sample is 0.5, possibly caused bythe light contribution from the AGN.Unsurprisingly, only a handful of bulgeless Sy2 galax-ies have estimates for M BH . Most of these observationsare confined to IR and X-rays measurements due to theobscuration present in Sy2 galaxies. One such example is NGC 3621, which was first discovered to have AGN ac-tivity through IR detections of [Ne V ] at 14 and 24 µ m(Satyapal et al. 2007). Subsequent observations of X-ray emission (Gliozzi et al. 2009) and stellar-dynamicalmodeling of the nuclear star cluster (Barth et al. 2009)have placed M BH between 4 × and 3 × M (cid:12) .[Ne V ] detection at 14 µ m was also detected in NGC4178 (Satyapal et al. 2009). Follow-up X-ray observa-tions (Secrest et al. 2012) indicate a M BH between 10 and 10 M (cid:12) . Shields et al. (2008) discovered a low-luminosity IMBH in NGC 1042 with an upper limit to M BH calculated at 3 × M (cid:12) based on the mass of idden BLR in Bulgeless Galaxy M BH in the rangeof 10 — 10 M (cid:12) and 10 — 10 M (cid:12) for NGC 3367and NGC 4536, respectively. Other bulgeless Sy2 galax-ies have only X-ray observations. These include NGC4561, which has a calculated lower mass limit of 2 × M (cid:12) (Araya Salvo et al. 2012), and NGC 3319, a barredgalaxy hosting an IMBH with an estimated upper limitof 3 × M (cid:12) (Jiang et al. 2018). With M BH ≈ . ,J0851+3928 is more massive (in some cases over an or-der of magnitude) than the other Sy2 bulgeless galaxieslisted here.In Figure 5, we plot M BH vs. M stellar for the bul-geless sample described above. Three additional bul-geless galaxies from Bentz & Katz (2015) and Daviset al. (2017) are included (see Tables 4 and 6 in Ap-pendix C for mass measurements). In addition, Rakshitet al. (2017) obtained H α line measurements of the NLS1galaxy NGC 3367, from which we calculated M BH us-ing the updated virial mass estimator from Woo et al.(2015) that incorporates the new value of log f = 0.05 ± f = 1.12), MM (cid:12) = 10 . ± . (cid:18) L H α erg s − (cid:19) . (cid:18) FWHM H α km s − (cid:19) . (3)References for stellar masses and all values are sum-marized in Appendix C and Table 3. If uncertainties arenot given, then they are assumed to be 0.3 dex.The line of best fit and confidence intervals were cal-culated using a Bayesian approach with linear regres-sion done by emcee (note that J0851+3928 was not in-cluded in this fit). A component of intrinsic scatter wasnot included due to the significant overlap of the largeerror bars. The best-fit line is weighed more heavilytoward the higher-mass BHs with dynamical mass esti-mates due to their smaller uncertainties. Fitting onlythe X-ray observations increased the uncertainty of theslope by almost a factor of 5. Because of the small sam-ple size and large uncertainties, a reliable fit is difficultto make. The virial estimate of M BH for J0851+3928puts it 0.77 dex below the relation but within the scat-ter of the other M BH with virial and dynamical massestimates. This indicates that the bulgeless BH massescalculated from X-rays are likely lower limits. This is notsurprising since many of these galaxies are Sy2, wheretheir high levels of extinction and large column densitiescan heavily obscure X-ray measurements. 4.2. M BH Relations
Simmons et al. (2013, 2017) have suggested thatSMBHs in disk-dominated galaxies are overmassive inthe M BH – M bulge relation and seem to outgrow theirbulge through secular processes unrelated to majormergers. In addition, they found that these SMBHsfollow the M BH – M stellar relation more closely. In orderto compare J0851+3928 to these results, we formed anextensive sample that incorporates a range of morpho-logical types with both Sy1 and Sy2 galaxies, includingthose with pseudobulges. The primary purpose here isto show how J0851+3928 compares to a large sampleof galaxies. The following sections describe the sampleand papers used, and all measurements are compiledin Appendix C. The full data set is also available fordownload.4.2.1. BH, Bulge, and Total Stellar Masses
The AGN BH Mass Database (Bentz & Katz 2015)provides a compilation of BH masses from reverberationmapping studies. The basic method of reverberationmapping is to monitor variations in the continuum fluxand broad emission lines and measure the light-traveltime delay between the two. M BH are derived from thesemeasurements using the virial relation given by Equa-tion (1). To properly compare the M BH of J0851+3926calculated using Equation (2), we need to adopt a con-sistent value of f . Because the reverberation masses inthe database are σ based, we use log f = 0.65 ± f = 4.47) as calibrated in Woo et al. (2015). M BH of 37galaxies with reliable bulge and stellar masses are listedin Table 4 (see Appendix C). The quoted errors includeuncertainties from both the database and from f , wheremost of the uncertainty arises.Graham & Scott (2015) compiled data from severaldifferent studies and selected the low-mass AGN whose M BH are undermassive relative to the M BH – M Bulge re-lation. These BH masses were calculated using single-epoch virial mass estimators that require the use of thevirial coefficient. We use the updated value of f ( f =1.12) (Woo et al. 2015), as was done in Sections 3.2and 4.1. Due to the need to recalculate all single-epochmass measurements, we only select AGN from Graham& Scott (2015) that have quoted emission-line measure-ments. The majority of the BH masses were calculatedusing Equation (3); however, for Pox 52, where λL Bohn et al. M Stellar (M ) M B H ( M ) BulgelessJ0851+3928
Figure 5. M BH plotted versus M stellar for purely bulgeless galaxies. J0851+3928 is represented as a blue star, and its errorbars for stellar mass are comparable to the size of the star symbol. The shaded contours are set at 1 σ confidence intervals, andthe dashed-dotted line represents the line of best fit (excluding J0851+3928). and FWHM H β are reported, we followed the relationderived in Sexton et al. (2019), MM (cid:12) = 10 . +0 . − . (cid:18) λ L erg s − (cid:19) . +0 . − . × (cid:18) FWHM H β km s − (cid:19) (4)which also incorporates the updated f value. Val-ues for M BH and the references of the measurementsare listed in Table 5 (see Appendix C). The errors re-ported for M BH include the quoted uncertainties in L H α , λL , FWHM H α , FWHM H β , and f .The rest of the BH masses in our sample were derivedusing dynamical mass measurements, including stellardynamics, gas dynamics, stellar orbit motions, and stim-ulated water maser emission. We compile M BH from thefollowing papers: all 44 late-type galaxies in Davis et al.(2017), 39 early-type galaxies from Sahu et al. (2019),37 galaxies from Savorgnan et al. (2016), and 3 galaxiesfrom Hu (2009). Masses and uncertainties are quotedfrom each paper and can be found in Table 6 (see Ap-pendix C).We also quote M bulge and M stellar values from the lit-erature. These values are listed in Tables 4–6 and arepredominately calculated from color-dependent stellarM/L ratios. The general procedure is to perform 2D bulge/disk decompositions while simultaneously fittingfor any structural features such as spiral arms, rings, andbars. Based on the surface brightness profiles, one canobtain apparent and absolute magnitudes from which lu-minosities can be estimated. With the appropriate M/Lratio, a mass for each component can be calculated. Asummary of the methods used to calculate bulge andtotal stellar mass in each referenced source is presentedin Appendix C. If uncertainties are not specified, thenthey are assumed to be 0.3 dex.Lastly, some quoted M bulge are greater than their hoststellar masses. Although this could naturally come fromthe use of different methods of fitting or different M/Lratios used, the mass discrepancy could also arise as aresult of color differences of the components. For ex-ample, in the case of late types in Bentz & Manne-Nicholas (2018), the bulge will tend to be redder andhave a higher V − H than the disk. A single V − H thatrepresents the entire galaxy will be bluer, which resultsin a M stellar that is less than M bulge . The differences,however, are typically within the quoted uncertainties.4.2.2. M BH - M bulge Relations
We first fit M BH and M bulge data described above us-ing emcee in a similar fashion to what was done inSection 4.1; however, a component of intrinsic scatterwas included in the fit. This was done since there is nota significant amount of overlap in the error bars. We fit idden BLR in Bulgeless Galaxy σ -confidenceintervals of the fit, and we find the best-fit relationshipto belog (cid:18) M BH M (cid:12) (cid:19) = (1 . ± . (cid:18) M bulge M (cid:12) (cid:19) +(7 . ± . y -intercept.We also investigated the morphological dependence ofthe sample and found that early-type galaxies have asteeper slope (1.13 ± ± < M bulge ≤ . M (cid:12) would indicate that J0851+3928hosts an overmassive BH compared to what the M BH - M bulge relation would predict: J0851+3928 is 1.59 dexabove the best-fit relation (Fig. 6, left panel), a factorof 2.84 above the scatter. Fitting only late types, wefind J0851+3928 to be 1.25 dex above the fit, a factor of2.40 above the scatter. For galaxies with similar M BH (within 10 . –10 . M (cid:12) ), the bulge mass of J0851+3928is at least 3.50 σ below the median.4.2.3. M BH - M stellar Relation
We used the same linear regression method to fit the M BH and M stellar data as was done for M BH – M bulge . The results of the fit to the entire data set are shown inthe right panel of Figure 6 and the best-fit line islog (cid:18) M BH M (cid:12) (cid:19) = (1 . ± . (cid:18) M stellar M (cid:12) (cid:19) +(6 . ± . M BH - M bulge .Like the M bulge relations, the various M stellar relationsof each subsample do not diverge beyond 1 dex of eachother within the low- and high-mass ends. The slope ofthe early types is steeper than late types, 1.24 ± ± M BH – M stellar relation (1.26 ± M BH – M bulge relation (1.15 ± M BH - M stellar is steeper is not surprisingsince the most massive galaxies hosting the most mas-sive BHs typically have higher B/T flux ratios (Daviset al. 2018). This causes galaxies with higher M bulge tobe shifted toward the right in the M BH – M bulge relation,thus lowering the steepness of the slope.Similar to the results found by Simmons et al. (2013,2017), J0851+3928 does fall closer to the M BH - M stellar ,differing by 0.97 dex, which is a factor of 1.28 belowthe relation. Also, it is well within the distribution ofthe late-type galaxies, the most morphologically similarsubsample, and only differs by 0.37 dex of the late-typebest fit (not shown).Also plotted in Figure 6 is the M BH - M stellar relationfor bulgeless galaxies from Figure 5. The bulgeless X-ray galaxies were not included in the full sample thatproduced Equation (6). We see a decrease in the slopeof the full sample, which is likely driven by the handfulof galaxies at the low-mass end that have higher M BH estimates than the X-ray sources. In addition, the offsetof the X-ray sources from the full sample further indi-cates that these are lower limits to the BH mass. Theother bulgeless galaxies with virial estimates for M BH ,including J0851+3928, fall closer to the relation, andJ0851+3928 falls well within the scatter of the otherlate type galaxies. The fact that all four of the bul-geless galaxies with robust estimates fall amongst bothearly and late-types suggests that perhaps the major BHgrowth mechanisms in bulgeless galaxies are not all thatdifferent. Although the sample size is still too small tomake any firm conclusions, it is certainly intriguing howthese BHs in bulgeless galaxies grew to supermassivesize without going through major merger events.4 Bohn et al. M Bulge (M ) M B H ( M ) Typical Uncertainty
Early-typeLenticularLate-typeJ0851+3928 M Stellar (M ) M B H ( M ) Typical Uncertainty
Early-typeLenticularLate-typeBulgelessJ0851+3928
Figure 6. M BH plotted versus M bulge (left) and M stellar (right). J0851+3928 is represented as a blue star, and the upper limitto M bulge is used here. Error bars for the stellar mass are comparable to the size of the star symbol. The shaded contours areset at 1 σ confidence intervals, and the black dashed-dotted line represents the line of best fit (excluding J0851+3928 and thebulgeless targets with only X-ray observations). The orange dashed line in the right panel represents the best-fit line to thebulgeless galaxies as is shown in Figure 5. The full list of individual values and uncertainties are found in Tables 4, 5 and 6 (seeAppendix C). Triggering of the AGN
The existence of a BH on the order of 10 . M (cid:12) ina galaxy with no obvious signatures of a major mergerraises the important question of how it has grown tosupermassive size. To trigger accretion, an inflow ofgas needs to be supplied to the central region. A natu-ral mechanism of this is a galaxy merger in which largequantities of gas can be sent toward the BH. The buildupof the bulge component is thought to accompany majormergers, so a different triggering mechanism likely trig-gered the AGN activity observed in J0851+3928. In thissection, we discuss two possible scenarios of how the cur-rent accretion onto the BH may have started: fly-by ofa companion galaxy and a galactic bar that can removeangular momentum from the gas.The SDSS postage stamp (top right panel of Fig-ure 2) shows a small galaxy about 9 (cid:48)(cid:48) away with somelow surface brightness emission potentially connectingit to J0851+3928. If this galaxy is indeed a companionrather than a close projection, a tidal interaction withJ0851+3928 could have disrupted the gas in the disk.If, in addition, this companion were gas-rich, some ofthe gas could have been accreted by the larger galaxy, possibly explaining the counterrotation observed in Sec-tion 3.5. In either case, the tidal interaction could haveprovided the means to remove angular momentum fromthe gas, thus funneling it onto the central engine.To investigate whether the small galaxy is a tidal com-panion, we obtained NIRES spectra to measure its red-shift (see Figure 4b for slit orientation). Unfortunately,the spectrum did not show any obvious emission or ab-sorption features in any of the NIRES bands. Thus, theonly redshift value available is the photometric redshift(PhotoZ) from SDSS. SDSS reports a PhotoZ = 0.257 ± ± ± × − ) forJ0851+3928. If this redshift is accurate, then the smallobject is a background galaxy rather than a close com-panion. This is consistent with the fact that we find noasymmetries in the residuals of the GALFIT decompo-sitions (see Figure 2), suggesting that the inner disk islargely intact and substantial interaction is unlikely.Many studies have shown that galactic bars are quitecommon in nearby spiral galaxies and may play a pivotalrole in the secular evolution of AGN (eg., Eskridge et al.2000; Jogee et al. 2005). Due to the nonaxisymmetric idden BLR in Bulgeless Galaxy NE5"
Figure 7.
Flux plot of J0851+3926. The contour level isarbitrarily set to bring out the spiral arm structure. distributions of mass, galactic bars may help drive gastoward the center through gravitational torques that re-duce the angular momentum of the gas, thus drivingit inward toward parsec scales (eg., Piner et al. 1995;Sheth et al. 2002; Wang et al. 2012). Bar formationcould have arisen from disk instabilities induced by aclose fly-by or cold gas accretion from dark matter fila-ments (Combes 2008). Through intersecting filaments,Algorry et al. (2014) have shown that a galactic barcan arise due to an inner counterrotating disk or bar.This can occur if accretion along the filaments occurs indifferent episodes, where the inner bar forms first, fol-lowed by the outer disk at a later time. Close inspectionof J0851+3928 reveals the possibility of a bar-like struc-ture. Figure 7 shows a contour, set at an arbitrary level,that emphasizes the spiral arms and possible bar struc-ture. In contrast, no clear bar-like feature is seen in the
GALFIT decompositions (see Figure 2). In order tobetter characterize the central region, higher-resolutiondata are needed to resolve the central kpc. CONCLUSIONWe have obtained NIRSPEC and NIRES NIRspectra and
Chandra
X-ray observations of SDSSJ085153.64+392611.76, a bulgeless, Sy2 galaxy withbroad Pa α emission. This offers us the special oppor-tunity to obtain a virial BH mass estimate while alsoallowing us to put strong constraints on any potentialgalactic bulge. Using virial mass estimators, we calcu-lated an extinction-corrected BH mass of log( M BH /M (cid:12) )= 6 . ± .
50. There is some ambiguity to the presenceof AGN activity in the SDSS spectrum that showcases that NIR selection techniques could be better suited inselecting and studying AGN that are deeply buried indust. Our lack of X-ray detection is consistent withthis scenario of a heavily obscured AGN and highlightsthe need for IR spectroscopic observations to uncoverhidden BHs in this demographic. Additionally, the lackof a bulge component in J0851+3926 indicates that itis unlikely to have undergone a major merger event andthat the central BH has grown to a supermassive sizequiescently. Clearly, some secular mechanism, likelyindependent from mergers, can fuel AGN and growSMBHs.We compiled a substantial sample of AGN, includ-ing those found in bulgeless galaxies, and find thatJ0851+3926 falls within the scatter of the M BH – M stellar relation. In addition, the virial mass estimate of the BHmass provides one of the most secure mass estimates of abulgeless Sy2 galaxy. Since it does not have a bulge com-ponent, we find that the M BH – M stellar relation is morereliable for bulgeless galaxies or those with pseudobulgesthan the M BH – M bulge relation. Obtaining total stellarmass of the galaxy is more straightforward than decon-volving the galaxy into individual components, particu-larly when the structures are not well resolved.We also report counterrotation of gas within the cen-tral kpc of J0851+3926. Possible causes of this include apotential faint bar that is changing the angular momen-tum of the gas or a close fly-by of a companion galaxythat disrupted the gas in the disk. Higher-resolution ob-servations will be needed to search for further evidenceof a bar and/or traces of tidal interactions.We thank the anonymous referee for their time andhelpful comments on this work. We also thank LisaPrato for her assistance with REDSPEC and GeorgeBecker for assisting with the NIR reduction pipeline.We thank Dr. Greg Doppmann and Dr. Percy Gomezfor supporting our Keck observations. We also thankDr. Laura Sales and Remington Sexton for insightfulconversations.Partial support for this project was provided by theNational Science Foundation, under grant No. AST1817233.Some of the data presented herein were obtained atthe W. M. Keck Observatory, which is operated as ascientific partnership among the California Institute ofTechnology, the University of California, and the Na-tional Aeronautics and Space Administration. The Ob-servatory was made possible by the generous financialsupport of the W. M. Keck Foundation.The authors wish to recognize and acknowledge thevery significant cultural role and reverence that the sum-6
Bohn et al.
Facilities:
Chandra, Keck:II (NIRSPEC, NIRES),Sloan.
Software:
BADASS (R.Sextonetal.2020,inprepara-tion, https://github.com/remingtonsexton/BADASS2),
CIAO (v4.11; software package (Fruscione et al. 2006)), emcee (Foreman-Mackeyetal.2013),
GALFIT (Pengetal.2010),
GANDALF (Sarzi et al. 2006), pPXF (Peng et al.2002, 2010),
PyRAF (PyRAF is a product of the SpaceTelescope Science Institute, which is operated by AURAforNASA),
REDSPEC A. NIRSPEC MEASUREMENTSHere we show the fit to the NIRSPEC data (see Fig. 8). As mentioned in Section 3.2, our NIRSPEC data have a highdegree of extinction due to heavy cloud cover, so we do not include the flux measurements in the bulk of our analysis.However, the FWHM of the broad component is still comparable to the NIRES data (see Table 2) and indicates thatthe broad emission is due to AGN activity and is not stellar in origin (see Section 3.2 for further details). B. SDSS MEASUREMENTSIn this section, we show the
BADASS (R. Sexton et al. 2020, in preparation) fits to the SDSS spectrum. We fittwo models: one with broad components included (Fig. 9, right panel) and the other without (Fig. 9, left panel). Thebest-fit model (overlaid in red) incorporates both Fe II emission and stellar absorption from the host galaxy. Outflowswere tested for, but none were detected (see R. Sexton et al. 2020, in preparation for further details). Although thefull SDSS spectrum was fit, we only show the H α complex since there is substantial absorption in H β and no broadH β emission is detected. The FWHM of the broad H α emission is 1911 . +248 . − . km s − , a factor of 1.4 above ourNIRES Pa α value. The amplitude is about two times the 1 σ level of the noise, suggesting that a component could bethere.To compare the fits and quantify whether a broad component is needed, we ran the F -test: F = ( σ single ) / ( σ double ) ,where σ is the standard deviation of the residuals using either single or double gaussian components. For the regionaround H α , we obtain F = 1.41. Although this value suggests that adding a broad component does improve the fit, wecannot say for certain whether a significant broad component exists; a value closer to 2 or 3 is needed to provide moreconclusive evidence. It is likely that the broad emission is heavily absorbed and any emission detected is dominatedby the noise in the spectrum. C. METHODS AND MEASUREMENTS FOR M BH RELATIONSBelow is a summary of the methods used to estimate M bulge and M stellar of our sample described in Sections 4.1 and4.2. The following tables quote the measurements and errors from each reference. Table 3 lists the bulgeless galaxies idden BLR in Bulgeless Galaxy F ( e r g c m s Å ) SpectrumBest FitPa Narrow ComponentPa Broad Component
Rest Wavelength (Angstroms) F R = 0.92 Figure 8.
The MCMC fit to the NIRSPEC spectrum. In the top panel, Pa α is centered with the best fit plotted over thespectrum. Below the spectrum are the narrow (dotted) and broad (solid) components. The bottom panel plots the residuals,1 σ noise level (horizontal dotted lines), and the R value. F ( e r g c m s Å ) SpectrumBest Fit
Rest Wavelength (Angstroms) F R = 0.96 SpectrumBest FitH Broad Component Rest Wavelength (Angstroms) R = 0.97 Figure 9.
The MCMC fit to the SDSS spectrum. In the two top panels, H α and [N II ] λλ α component, plotted in green. The bottom two panels plot the residuals, the1 σ noise level (horizontal dotted lines), and the R values. Bohn et al. discussed in Section 4.1. The subsequent tables list mass measurements from Section 4.2 and are categorized basedon the method used to estimate M BH . Note that the M BH derived from reverberation mapping and the virial methodwere recalculated using the updated f factor from Woo et al. (2015) (see Section 4.2.1). Lastly, if uncertainties werenot listed, then they are assumed to be 0.3 dex. All of the following tables are available for download.Fall & Romanowsky (2018) calculated total stellar masses from K -band (2.2 µ m) luminosities using a mass-to-lightratio, M ∗ / L K , based on B − V colors. To estimate M/L, they used M ∗ / L K = 0.96( B − V ) + 0.01. For spiral galaxies,which NGC 1024 is classified as, disk stellar masses were calculated separately and summed together with any measuredbulge component to get the total stellar mass.Davis et al. (2018, 2019) performed 2D decompositions of 3.6 µ m images from the Spitzer Survey of Stellar Structurein Galaxies, with additional imaging from the Hubble Space Telescope (HST)
F814W filter and the Two Micron AllSky Survey (2MASS) K s band (2.2 µ m). M/L ratios of 0.60, 1.88, and 0.62 were used for Spitzer , HST , and 2MASSdata. To account for the contribution of dust emission at 3.6 µ m, a ∼
25% reduction in the luminosity was included,leading to a M ∗ /L obs , IRAC1 of 0.453 for dusty galaxies.Georgiev et al. (2016) obtained total stellar masses using the M/L color relations derived in Bell et al. (2003). B magnitudes and B − V colors were obtained from HyperLeda and were used in log( M ∗ /L B ) = 1.737( B − V ) - 0.942to obtain M stellar for NGC 3319.Kelly & Kirshner (2012) calculated total stellar masses by first estimating the flux using MAG AUTO in Sextrac-tor , where the radius of the aperture was set to be 2.5 Kron. M/L ratios were then estimated with SED fitting using PEGASE2 stellar population models.McGaugh & Schombert (2014) compiled mass-to-light relations for a number of wavelength bands ( V , I , and 3.6 µ m) from various sources, including those from Bell et al. (2003) and Into & Portinari (2013), where B − V colorswere used. Magnitudes were taken from Spitzer and 2MASS. Values of M stellar for NGC 3621 are consistent across allbands (within the assumed error of 0.3 dex), so we use the average of the values given by Into & Portinari (2013).Hughes et al. (2013) utilized the B − V color-dependent relations derived in Bell et al. (2003). H -band (1.65 µ m)luminosities were taken from 2MASS and used in the equation: log( M ∗ /L H ) = 0.21( B − V ) – 0.059. B − V colors werecalculated using either B and V magnitudes from GOLDMine or morphologically averaged values if observationswere not available (taken from NED ).Bentz & Manne-Nicholas (2018) obtained both optical and NIR imaging of their sample, with the optical datacoming from high-resolution medium-band V HST observations and the NIR data from H -band images taken atWIYN Observatory. 2D decompositions were done with GALFIT with the higher-resolution HST fits guiding theNIR fit parameters. Masses were calculated using V − H colors and the M/L relation M ∗ /L V = 1.493( V − H ) – 0.681as derived in Into & Portinari (2013).Reines & Volonteri (2015) present M stellar of a sample of broad-line AGN utilizing SDSS g - and i -band photometry.A mock AGN spectrum was constructed for each source and then removed to isolate the host luminosity contribution.After correcting for galactic reddening, host galaxy masses were then calculated using log( M ∗ /L i ) = 1.032( g − i ) –0.963 (Zibetti et al. 2009). Additional stellar masses for a sample of dwarf galaxies, galaxies with reverberation-mappedAGN, and galaxies with dynamical M BH are also provided. The AGN contribution was removed from the dwarf galaxyand reverberation-mapped subsamples, and the stellar mass was obtained in the same way as for the broad-line AGN.For the dynamical BH mass sample and Pox 52, which is not in the SDSS footprint, B and V magnitudes were obtainedand stellar masses were calculated using log( M ∗ /L K ) = 1.176( B − V ) – 1.390 (Zibetti et al. 2009). For Pox 52, a dwarfelliptical, Barth et al. (2004) did not find any indication of a spiral or disk component with GALFIT decompositions,and so we adopt the same value of M stellar as M bulge .For UM 625, Graham & Scott (2015) report a M bulge of 5.4 × M (cid:12) . This is estimated from a V -band bulgemagnitude of -19.06 and stellar M/L ratio of 1.6 (Jiang et al. 2013). For M stellar , we use the value given by Stern &Laor (2013), who obtained masses from SDSS z -band photometry. After removal of the AGN contribution, luminositieswere converted to masses through a L [OIII] -dependent M/L ratio. Ratios ranged from 2.6 to 1.7 based on luminositiesbetween 10 and 10 . erg s − . http://leda.univ-lyon1.fr/ http://astroa.physics.metu.edu.tr/MANUALS/sextractor/ http://goldmine.mib.infn.it/ https://ned.ipac.caltech.edu/ idden BLR in Bulgeless Galaxy r -band B/Tgreater than 0.5 (0.53 and 0.58, respectively, when using a de Vaucouleurs model). Using the stellar mass and assuminga constant M/L ratio, we can obtain a rough estimate for M bulge : log M bulge = 9.31 ± M bulge = 9.50 ± M bulge of their active dwarfgalaxy sample. Optical and IR HST images were run through GALFIT to acquire magnitudes of each component.Magnitudes from HST filters (F606W and F110W) were then converted into SDSS r and g and 2MASS J magnitudesby fitting a wavelength-dependent flux density power law to the HST measurements and then evaluating the fit atthe appropriate wavelengths. These new magnitudes were subsequently used in the M/L relation log( M ∗ /L J ) =1.398( r − z ) – 1.271 provided by Zibetti et al. (2009).Savorgnan et al. (2016) report bulge luminosities derived from decompositions of 3.6 µ m Spitzer images. IndividualM/L ratios based on [3.6] – [4.5] colors were used in the relation log( M ∗ /L . ) = 3.98( ± ± K -band images from 2MASS and ran them through BUDDA , a 2D decomposition program,to obtain bulge luminosities. Masses of the bulges were then calculated from either log( M ∗ /L K ) = 0.135( B − V ) –0.356 or log( M ∗ /L K ) = 0.349( r − i ) – 0.336 (Bell et al. 2003) where extinction-corrected B − V colors are providedby HyperLeda and the r − i colors are from SDSS. If an AGN component was detected, the central 3 (cid:48)(cid:48) region wasremoved to avoid contamination from the AGN. When available, we choose the r − i relation since the magnitudes ofthe bulge effective radius were directly measured using the SDSS images.Sahu et al. (2019) provide decompositions of early-type galaxies with archived Spitzer
IRAC 3.6 µ m images, SDSS r -band images, or 2MASS K s -band images. PROFILER (Ciambur 2015, 2016) and two
IRAF tasks,
ISOFIT and
CMODEL , were used to model individual galaxy components and obtain magnitudes from which luminosities for theentire galaxy and bulge could be calculated. These luminosities were converted to stellar masses using the followingconstant stellar M/L ratios for each band: M ∗ /L . µm = 0.6, M ∗ /L K s = 0.7, and M ∗ /L r = 2.8.For three galaxies, NGC 3414, NGC 4621, and NGC 5846, M stellar were obtained from Dabringhausen & Fellhauer(2016). Here, age, color, and luminosity were all treated as parameters in their M/L calculations. Ages of the stellarpopulation for all three galaxies are quoted from McDermid et al. (2015). V -band luminosity and B − V colors camefrom HyperLeda , and g − r and g − i values came from SDSS. A M/L ratio and M stellar were then calculated fromthese values (see Equation 18 and Table 13 in Dabringhausen & Fellhauer (2016)). ∼ dgadotti/budda.html Bohn et al.
Table 3 . Bulgeless Galaxy SampleGalaxy log( M BH M (cid:12) ) Ref log( M stellar M (cid:12) ) RefNGC 1024 1.78 - 6.48 1 11.21 ± ± ± ± ± ± ± ± > a
14 9.63 ± ± a Lower Limit
Note —Columns: (1) Galaxy name. (2) M BH estimates from X-ray, IR,or virial measurements, typically lower/upper limits. (3) References for M BH . (4) Estimates for M stellar of the galaxy. (5) References for M stellar .References: (1) (Shields et al. 2008). (2) (Fall & Romanowsky 2018). (3)(Davis et al. 2017). (4) (Davis et al. 2018). (5) (Jiang et al. 2018). (6)(Georgiev et al. 2016). (7) (Rakshit et al. 2017). (8) (Kelly & Kirshner2012). (9) (Satyapal et al. 2007; Barth et al. 2009; Gliozzi et al. 2009).(10) (McGaugh & Schombert 2014). (11) (Secrest et al. 2012). (12)(Hughes et al. 2013). (13) (McAlpine et al. 2011). (14) (Araya Salvoet al. 2012). Table 4 . Galaxy Sample with Reverberation Mapped M BH MeasurementsGalaxy log( M BH M (cid:12) ) Ref log( M bulge M (cid:12) ) Ref log( M stellar M (cid:12) ) RefArk 120 8.08 (+0.17, -0.18) 1 10.53 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Table 4 continued idden BLR in Bulgeless Galaxy Table 4 (continued)
Galaxy log( M BH M (cid:12) ) Ref log( M bulge M (cid:12) ) Ref log( M stellar M (cid:12) ) RefPG 1700+518 8.80 (+0.21, -0.22) 1 10.69 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Note —Columns: (1) Galaxy name. (2) M BH estimates from reverberation mapping. (3) Refer-ences for M BH . (4) Estimates for M bulge . (5) References for M bulge . (6) Estimates for M stellar of the galaxy. (7) References for M stellar .References: (1) (Bentz & Katz 2015). (2) (Bentz & Manne-Nicholas 2018). (3) (Davis et al.2019). (4) (Davis et al. 2018). Table 5 . Galaxy Sample with Virial M BH MeasurementsGalaxy FWHM H α a log(L H α ) b log( M BH M (cid:12) ) Ref log( M bulge M (cid:12) ) Ref log( M stellar M (cid:12) ) RefPox 52 765 ± c d ± ± ±
48 40.36 ± ± ± ±
224 39.53 ± ± ± ±
145 39.81 ± ± ± ±
158 39.38 (+0.06, -0.08) 5.81 (+0.22, -0.27) 4 8.21 ± ± ±
70 40.15 ± ± ± Table 5 continued Bohn et al.
Table 5 (continued)
Galaxy FWHM H α a log(L H α ) b log( M BH M (cid:12) ) Ref log( M bulge M (cid:12) ) Ref log( M stellar M (cid:12) ) RefSDSS J095418.15+471725.1 636 ±
64 39.41 ± ± ± ±
75 39.73 (+0.05, -0.06) 5.31 (+0.21, -0.26) 4 8.14 ± ± ±
89 39.67 (+0.05, -0.06) 5.44 (+0.21, -0.26) 4 7.87 ± ± ±
104 40.16 ± ± ± ±
79 39.45 ± ± ± a FWHM are in units of km s − . b Luminosities are in units of erg s − . c FWHM of H β . d log( λ L ). Note —Columns: (1) Galaxy Name. (2) FWM of the broad H α (or H β for Pox 52) emission line. (3) Luminosity of broad H α (or λ L for Pox 52). (3) M BH estimates from virial measurements. (3) References for M BH . (4) Estimates for M bulge . (5)References for M bulge . (6) Estimates for M stellar of the galaxy. (7) References for M stellar .References: (1) (Thornton et al. 2008). (2) (Jiang et al. 2013). (3) (Yuan et al. 2014). (4) (Reines et al. 2013). (5) (Reines &Volonteri 2015). (6) (Graham & Scott 2015). (7) (Omand et al. 2014) (8) (Schutte et al. 2019). (9) (Stern & Laor 2013). (10)(Chang et al. 2015). Table 6 . Galaxy Sample with Dynamical M BH MeasurementsGalaxy log( M BH M (cid:12) ) Ref log( M bulge M (cid:12) ) Ref log( M stellar M (cid:12) ) RefMilky Way 6.60 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Table 6 continued idden BLR in Bulgeless Galaxy Table 6 (continued)
Galaxy log( M BH M (cid:12) ) Ref log( M bulge M (cid:12) ) Ref log( M stellar M (cid:12) ) RefNGC 1097 8.38 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Table 6 continued Bohn et al.
Table 6 (continued)
Galaxy log( M BH M (cid:12) ) Ref log( M bulge M (cid:12) ) Ref log( M stellar M (cid:12) ) RefNGC 4026 8.26 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Table 6 continued idden BLR in Bulgeless Galaxy Table 6 (continued)
Galaxy log( M BH M (cid:12) ) Ref log( M bulge M (cid:12) ) Ref log( M stellar M (cid:12) ) RefNGC 5077 8.87 (+0.21, -0.23) 2 11.03 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± Note —Columns: Same as table 4 but with M BH estimates from dynamical measurements.References: (1) (Davis et al. 2017). (2) (Savorgnan et al. 2016). (3) (Hu 2009). (4) (Sahu et al. 2019). (5) (Daviset al. 2019). (6) (Davis et al. 2018). (7) (Reines & Volonteri 2015). (8) (Dabringhausen & Fellhauer 2016). REFERENCES
Algorry, D. G., Navarro, J. F., Abadi, M. G., et al. 2014,MNRAS, 437, 3596, doi: 10.1093/mnras/stt2154Antonucci, R. R. J., & Miller, J. S. 1985, ApJ, 297, 621,doi: 10.1086/163559Araya Salvo, C., Mathur, S., Ghosh, H., Fiore, F., &Ferrarese, L. 2012, ApJ, 757, 179,doi: 10.1088/0004-637X/757/2/179Assef, R. J., Stern, D., Kochanek, C. S., et al. 2013, ApJ,772, 26, doi: 10.1088/0004-637X/772/1/26Baldassare, V. F., Reines, A. E., Gallo, E., et al. 2016, ApJ,829, 57, doi: 10.3847/0004-637X/829/1/57 Baldwin, J. A., Phillips, M. M., & Terlevich, R. 1981,PASP, 93, 5, doi: 10.1086/130766Barth, A. J., Ho, L. C., Rutledge, R. E., & Sargent, W.L. W. 2004, ApJ, 607, 90, doi: 10.1086/383302Barth, A. J., Strigari, L. E., Bentz, M. C., Greene, J. E., &Ho, L. C. 2009, ApJ, 690, 1031,doi: 10.1088/0004-637X/690/1/1031Becker, G. D., Rauch, M., & Sargent, W. L. W. 2009, ApJ,698, 1010, doi: 10.1088/0004-637X/698/2/1010Bell, E. F., McIntosh, D. H., Katz, N., & Weinberg, M. D.2003, ApJS, 149, 289, doi: 10.1086/378847 Bohn et al.